Handle mod on ambiguous AA range + small fixes

Author: douweschulte

mzcore/src/ontology/ontologies.rs

@@ -3,7 +3,7 @@
 use std::sync::LazyLock;
 
 use context_error::*;
-use mzcv::{CVIndex, CVStructure, CVVersion, ControlledVocabulary};
+use mzcv::{AccessionCode, CVIndex, CVStructure, CVVersion, ControlledVocabulary};
 use serde::{Deserialize, Serialize};
 
 use crate::{
@@ -30,7 +30,7 @@ pub static STATIC_ONTOLOGIES: LazyLock<Ontologies> = LazyLock::new(Ontologies::i
 /// let modification = ontologies.get_by_name(&[], "Oxidation").unwrap();
 /// assert_eq!(modification.formula(), molecular_formula!(O 1));
 /// // or by index from a particular ontology
-/// let modification2 = ontologies.unimod().get_by_index(&35).unwrap();
+/// let modification2 = ontologies.unimod().get_by_index(&mzcv::AccessionCode::Numeric(35)).unwrap();
 /// assert_eq!(modification, modification2);
 /// // or search all (or a subset) for fuzzy matches
 /// let search = ontologies.search(&[], "Oxidated");
@@ -336,6 +336,48 @@ impl Ontologies {
         None
     }
 
+    /// Find the given index in this ontology.
+    pub fn get_by_index(
+        &self,
+        ontology: Ontology,
+        id: &AccessionCode,
+    ) -> Option<SimpleModification> {
+        match ontology {
+            Ontology::Unimod => {
+                if let Some(m) = self.unimod.get_by_index(id) {
+                    return Some(m);
+                }
+            }
+            Ontology::Psimod => {
+                if let Some(m) = self.psimod.get_by_index(id) {
+                    return Some(m);
+                }
+            }
+            Ontology::Xlmod => {
+                if let Some(m) = self.xlmod.get_by_index(id) {
+                    return Some(m);
+                }
+            }
+            Ontology::Gnome => {
+                if let Some(m) = self.gnome.get_by_index(id) {
+                    return Some(m);
+                }
+            }
+            Ontology::Resid => {
+                if let Some(m) = self.resid.get_by_index(id) {
+                    return Some(m);
+                }
+            }
+            Ontology::Custom => {
+                if let Some(m) = self.custom.get_by_index(id) {
+                    return Some(m);
+                }
+            }
+        }
+
+        None
+    }
+
     /// Get all modifications in the selected ontologies (or in all if the list is empty).
     pub fn data(&self, ontologies: &[Ontology]) -> impl Iterator<Item = SimpleModification> {
         let ontologies = if ontologies.is_empty() {

mzcore/src/sequence/peptidoform/parse.rs

@@ -632,70 +632,71 @@ impl PeptidoformIonSet {
         while index < chars.len() && index < range.end {
             match (c_term, chars[index]) {
                 (false, b'(') if chars.get(index + 1) == Some(&b'?') => {
-                    if braces_start.is_some() {
+                    if ambiguous_aa.is_some() {
                         combine_error(
                             &mut errors,
                             BoxedError::new(
                                 BasicKind::Error,
                                 "Invalid ambiguous amino acid set",
-                                "Ambiguous amino acid sets cannot be nested within ranged ambiguous modifications",
+                                "Ambiguous amino acid sets cannot be nested within ambiguous amino acid sets",
                                 base_context.clone().add_highlight((0, index, 1)),
                             ),
                         );
                         return Err(errors);
-                    }
-                    if ambiguous_aa.is_some() {
+                    } else if braces_start.is_some() {
                         combine_error(
                             &mut errors,
                             BoxedError::new(
                                 BasicKind::Error,
                                 "Invalid ambiguous amino acid set",
-                                "Ambiguous amino acid sets cannot be nested within ambiguous amino acid sets",
+                                "Ambiguous amino acid sets cannot be nested within ranged ambiguous modifications",
                                 base_context.clone().add_highlight((0, index, 1)),
                             ),
                         );
                         return Err(errors);
                     }
+
                     ambiguous_aa = Some(ambiguous_aa_counter);
+                    braces_start = Some(peptide.len());
                     ambiguous_aa_counter = handle!(single errors, ambiguous_aa_counter.checked_add(1).ok_or_else(|| {BoxedError::new(BasicKind::Error,
                         "Invalid ambiguous amino acid set",
-                        format!("There are too many ambiguous amino acid sets, there can only be {} in one linear peptide", std::num::NonZeroU32::MAX),
+                        format!("There are too many ambiguous amino acid sets, there can only be {} in one linear peptidoform", std::num::NonZeroU32::MAX),
                         base_context.clone().add_highlight((0, index, 1)))}));
                     index += 2;
                 }
-                (false, b')') if ambiguous_aa.is_some() => {
-                    ambiguous_aa = None;
-                    index += 1;
-                }
                 (false, b'(') => {
-                    if braces_start.is_some() {
+                    if ambiguous_aa.is_some() {
                         combine_error(
                             &mut errors,
                             BoxedError::new(
                                 BasicKind::Error,
                                 "Invalid ranged ambiguous modification",
-                                "Ranged ambiguous modifications cannot be nested within ranged ambiguous modifications",
+                                "Ranged ambiguous modifications cannot be nested within ambiguous amino acid sets",
                                 base_context.clone().add_highlight((0, index, 1)),
                             ),
                         );
                         return Err(errors);
-                    }
-                    if ambiguous_aa.is_some() {
+                    } else if braces_start.is_some() {
                         combine_error(
                             &mut errors,
                             BoxedError::new(
                                 BasicKind::Error,
                                 "Invalid ranged ambiguous modification",
-                                "Ranged ambiguous modifications cannot be nested within ambiguous amino acid sets",
+                                "Ranged ambiguous modifications cannot be nested within ranged ambiguous modifications",
                                 base_context.clone().add_highlight((0, index, 1)),
                             ),
                         );
                         return Err(errors);
                     }
+
                     braces_start = Some(peptide.len());
                     index += 1;
                 }
                 (false, b')') if braces_start.is_some() => {
+                    if ambiguous_aa.is_some() {
+                        ambiguous_aa = None;
+                    }
+
                     let start = braces_start.unwrap();
                     if start == peptide.len() {
                         combine_error(
@@ -981,6 +982,7 @@ impl PeptidoformIonSet {
             );
         }
 
+        dbg!(&ranged_unknown_position_modifications);
         if let Err(errs) = peptide.apply_ranged_unknown_position_modification(
             &ranged_unknown_position_modifications,
             &ambiguous_lookup,

mzcore/src/sequence/peptidoform/tests/positive.rs

@@ -13,7 +13,7 @@ Id,Example,Source,Key,Notes
 11,"EM[Oxidation]EVE[Cation:Mg[II]]ES[Phospho]PEK",spec2_0,"",""
 12,"EM[MOD:00719]EVEES[MOD:00046]PEK",spec2_0,"",""
 13,"EM[UNIMOD:35]EVEES[UNIMOD:56]PEK",spec2_0,"",""
-14,"EM[RESID:AA0581]EVEES[RESID:AA0037]PEK",spec2_0,"",""
+14,"EM[RESID:AA0581]EVEES[RESID:AA0037]PEK",spec2_0,"casing_specific",""
 15,"EMEVTK[XLMOD:02001#XL1]SESPEK[#XL1]",spec2_0,"",""
 16,"EMK[XLMOD:02000#XL1]EVTKSE[XLMOD:02010#XL2]SK[#XL1]PEK[#XL2]AR",spec2_0,"",""
 17,"EMEVTK[XLMOD:02001#XL1]SESPEK",spec2_0,"",""
@@ -137,7 +137,7 @@ Id,Example,Source,Key,Notes
 135,"PEPM[U:Oxidation]AS[M:O-phospho-L-serine]",spec2_1,"",""
 136,"{Phospho#g1}EM[U:Oxidation]EVS[#g1]",spec2_1,"ignore",""
 137,"{Sulfo#s}HT[#s]T[#s]S[#s]FRGC[#s]",spec2_1,"ignore",""
-138,"EM[R:L-methionine sulfone]EM[RESID:AA0581]",spec2_1,"",""
+138,"EM[R:L-methionine sulfone]EM[RESID:AA0581]",spec2_1,"casing_specific",""
 139,"(>Heavy chain)EVQLVESG",spec2_1,"casing_specific",""
 140,"EVTK[X:formaldehyde]LEK[XLMOD:02041]SEF",spec2_1,"",""
 141,"EVTK[X:formaldehyde#XL1]LEK[#XL1]SEF",spec2_1,"",""
@@ -246,7 +246,7 @@ Id,Example,Source,Key,Notes
 244,"{HexNAc#g1}PEPN[#g1]ITE",spec2_1,"ignore",""
 245,"{Hex#g1}{Sulfo#s}[#g1]-PN[#g1]KS[#s]C[#s]M",spec2_1,"ignore",""
 246,"{Glycan:Hex5HexNAc5#g1}PEPSTAT[#g1]IS[#g1]T[#g1]ICS[#g1]S[#g1]T[#g1]RIKES[#g1]IT[#g1]ES[#g1]",spec2_1,"ignore",""
-247,"EM[RESID:AA0581]EVEES[RESID:AA0037]PEK",spec2_1,"",""
+247,"EM[RESID:AA0581]EVEES[RESID:AA0037]PEK",spec2_1,"casing_specific",""
 248,"(>My (very) nice peptide)PEPTIDE", spec2_1, "casing_specific", ""
 249,"(>Interesting ]][ yes | [valid] #description_not_tag)PEPTIDE", spec2_1, "casing_specific", ""
 250,"(>Trypsin)AANSIPYQVSLNS+(>Keratin)AKEQFERQTA",spec2_1,"casing_specific",""
@@ -368,7 +368,7 @@ Id,Example,Source,Key,Notes
 366,"(PEP)[Oxidation#1(0.95)]M[#1(0.05)]AT",example,"just_parse",""
 367,"(?VCH)AT",example,"",""
 368,"PEPM[U:Oxidation]AS[M:O-phospho-L-serine]",example,"",""
-369,"EM[R:L-methionine sulfone]EM[RESID:AA0581]",example,"",""
+369,"EM[R:L-methionine sulfone]EM[RESID:AA0581]",example,"casing_specific",""
 370,"(>Heavy chain)EVQLVESG",example,"casing_specific",""
 371,"ED[MOD:00093#BRANCH]//D[#BRANCH]ATR",example,"casing_specific",""
 372,"NEEYN[Glycan:Hex5HexNAc4NeuAc1]K",example,"",""

mzcv/src/curie.rs

@@ -1286,6 +1286,12 @@ pub enum AccessionCode {
     //Big(ThinStr),
 }
 
+impl From<u32> for AccessionCode {
+    fn from(value: u32) -> Self {
+        Self::Numeric(value)
+    }
+}
+
 impl Default for AccessionCode {
     fn default() -> Self {
         Self::Numeric(0)

mzcv/src/cv_source.rs

@@ -20,11 +20,11 @@ use crate::{CVError, Curie, hash_buf_reader::HashBufReader};
 /// }
 /// ```
 pub trait CVSource {
-    /// Set this constant to true to enable automatic writing of the CV when the cahe is updated
+    /// Set this constant to true to enable automatic writing of the CV when the cache is updated
     const AUTOMATICALLY_WRITE_UNCOMPRESSED: bool = false;
     /// The data item that is stored in the CV
     type Data: CVData + 'static;
-    /// The type of the main datastructure to keep all data items (used to build any kind of hierarchy necessary)
+    /// The type of the main data structure to keep all data items (used to build any kind of hierarchy necessary)
     type Structure: CVStructure<Self::Data> + Encode + Decode<()>;
     /// The name of the CV, used to create the paths to store intermediate files and caches so has to be valid in that context
     fn cv_name() -> &'static str;
@@ -48,7 +48,7 @@ pub trait CVSource {
         folder.join(Self::cv_name())
     }
     /// Parse the textual representation of this CV. Return the version, the data, and possibly a
-    /// list of warnings or non critical errors encountered while parsing.
+    /// list of warnings or noncritical errors encountered while parsing.
     /// # Errors
     /// If the parsing failed.
     fn parse(
@@ -183,7 +183,7 @@ pub trait CVStructure<Data>: Default {
         Self: 'a;
     /// Iterate over all data items
     fn iter_data(&self) -> Self::IterData<'_>;
-    /// Add a sinlge data item to the structure
+    /// Add a single data item to the structure
     fn add(&mut self, data: std::sync::Arc<Data>);
     /// The indexing type
     type Index: Clone;